nature publishing group
Translational Investigation
Articles
Do children with congenital hypothyroidism exhibit abnormal cortical morphology? Hayyah Clairman1, Jovanka Skocic1, Julieta E. Lischinsky2 and Joanne Rovet1,3 Background: Given thyroid hormone (TH)’s essential role in multiple aspects of early brain development, children with congenital hypothyroidism (CH) detected and treated early may still display subtle cognitive and behavioral impairments as well as brain abnormalities. However, effects on their cortical development are not yet known. We used an automated neuroimaging technique to determine if these children differ in cortical thickness (CT) from typically developing controls (TDC) and if the regions showing CT differences reflect severity of initial hypothyroidism and predict later neuropsychological functioning. Methods: FreeSurfer Image Analysis Suite was used on archived MRI scans from 41 CH and 42 TDC children aged 9–16 y. Vertex-based procedures were used to compare groups and perform correlations between CT and indices of disease severity and neuropsychological outcome. Results: The CH group showed multiple regions of cortical thinning or cortical thickening within right and left hemispheres relative to TDC. CT values were significantly correlated with early T4 and thyroid-stimulating hormone (TSH) levels and current neuropsychological test indices. Conclusion: The developing cortex is sensitive to early TH loss in CH. Different patterns of cortical thinning or cortical thickening among brain regions may reflect timing of TH deficiency relative to timing of cortical development.
C
ongenital hypothyroidism (CH) is a pediatric endocrine disorder that affects ~1 in 2,500 newborns (1) and is caused by a perinatal lack of thyroid hormone (TH). CH arises from abnormal development or function of the thyroid gland that produces TH or in a small proportion of cases, a defect in the hypothalamus or pituitary, which regulate the thyroid (2). Because TH is critical for many aspects of early brain development, CH was formerly a leading cause of mental retardation due to its long period before diagnosis and treatment (3). Since the advent of newborn screening, most affected yet asymptomatic children are now diagnosed and treated right after birth. However, they are still at risk for suboptimal development due to their brief period of TH deficiency. A considerable body of research on these children has shown that they
have reduced IQ (3), poor school performance (4), behavior problems (5), and mild to moderate cognitive difficulties in language, visuospatial, sensorimotor, memory, and attention areas (6). Rodent models of early hypothyroidism have been instrumental in pinpointing the specific manifestations of early TH deficiency within the developing brain. When TH is missing, fundamental neurobiological processes such as neurogenesis, neuronal migration, synaptogenesis, and myelination develop atypically (7), while structures such as hippocampus (8), striatum (9), and cortex (10,11) are abnormal. Within the cortex, a late gestational/early postnatal TH insufficiency is associated with disturbed corticogenesis due to abnormal asymmetric division of neurons migrating toward the cortical surface (12). Since this aspect of corticogenesis gives rise to the particular cytoarchitectural feature known as cortical thickness (CT), it is possible that this may also be abnormal in children with CH. Advances in neuroimaging now make it possible to evaluate human cortical morphology. A commonly used tool is FreeSurfer Image Analysis Suite (Laboratory for Computational Neuroimaging, Martinos Center for Biomedical Imaging, Charlestown, MA), which via a set of automated algorithms, precisely reproduces gray/white and pial surfaces and measures CT by the distance between corresponding vertices on both surfaces (13). Research using this technique has shown distinct CT abnormalities in various pediatric conditions including autism (14), prematurity (15), attention deficit hyperactivity disorder (ADHD) (16), and fetal alcohol spectrum disorder (17). Previously, we reported that children with CH show structural and/or functional abnormalities of the hippocampus (18,19) but their cortex was not investigated. Therefore, we performed FreeSurfer (Laboratory for Computational Neuroimaging, Martinos Center for Biomedical Imaging) analyses on archived MRI scans of CH and typically developing control (TDC) groups. Our goals were to (i) identify brain regions where groups differed in CT, (ii) evaluate the relationships between patterns of cortical thinning or thickening and CH severity, and (iii) examine the relationships between CT and cognitive abilities in CH. We hypothesized that the CH group will exhibit distinct CT abnormalities that will be
1 Neurosciences and Mental Health Program, The Hospital for Sick Children, Toronto, Ontario, Canada; 2Institute for Biomedical Sciences, The George Washington University, Washington DC; 3Department of Pediatrics, University of Toronto, Toronto, Ontario, Canada. Correspondence: Joanne Rovet (
[email protected])
Received 13 August 2014; accepted 3 February 2015; advance online publication 17 June 2015. doi:10.1038/pr.2015.93 Copyright © 2015 International Pediatric Research Foundation, Inc.
Pediatric Research 1
Articles
Clairman et al.
associated both with severity of initial hypothyroidism and current weaknesses in neuropsychological functioning. RESULTS Demographic, Neuropsychological, and Biomedical Results
Table 1 presents demographic and neuropsychological test findings of CH and TDC groups. Groups did not differ in age, sex, handedness, socioeconomic status, or intracranial, gray matter, white matter, or cerebrospinal fluid volumes. CH scored significantly below TDC on the Wechsler Abbreviated Scale of Intelligence (WASI) Full Scale IQ, Vocabulary, and Matrix Reasoning (20) but not on the Rey-Osterrieth Complex Figure Test (REY-O) (21). Table 2 contains biomedical data for the CH group. Findings revealed a mean ± SD thyroxine value at diagnosis (T4dx) of 58.9 ± 39.4 nmol/l, which is in the hypothyroid range (normal range = 65.0–165 nmol/l). Their thyroid stimulating hormone levels at diagnosis (TSHdx) were well above normal (median = 265.1 mU/l; normal range = 0.5–5.0 mU/l) and Table 1. Mean (± SD) demographic and neuropsychological test data for CH and control groups CH group (n = 42) Age (years)
12.4 ± 1.8
Sex (males)
Control group (n = 42)
20
Socioeconomic status levela
ns
20
1.9 ± 0.92
ns ns
2.0 ± 0.94
83
Handedness (% RH)b
P value
12.0 ± 1.6
88
ns
Intracranial volume
1,709.6 ± 179.0c
1,735.8 ± 188.0
ns
Gray matter volume
847.8 ± 96.0c
869.4 ± 84.5
ns
White matter volume
428.0 ± 56.8c
420.3 ± 45.4
ns
Cerebrospinal fluid volume
433.8 ± 98.7c
446.1 ± 131.5
ns
IQ
105.5 ± 10.1
115.8 ± 9.8
0.000
WASI vocabulary
10.8 ± 2.3
13.2 ± 2.5
0.000
WASI matrix reasoning
11.1 ± 1.8
12.2 ± 1.6
0.004
−0.421 ± 1.1
−0.178 ± 1.0
ns
−0.980 ± 1.2
−0.538 ± 1.0
ns
REY-O copyd REY-O delayed
d
REY-O, Rey-Osterrieth Complex Figure Test; WASI, Wechsler Abbreviated Scale of Intelligence. a Range = 1 (highest) to 5 (lowest). bHandedness of one CH and two controls were unknown. cn = 41 as one CH scan was excluded from analyses due to excessive motion. d CH group, n = 41; control group, n = 38.
Table 2. Congenital hypothyroidism group’s biomedical data at time of diagnosis Value T4dx (nmol/l)a TSHdx (mU/l)b Age (days) at start of treatmenta
58.9 ± 39.4 265.1 16.8 ± 13.3
Starting dose (µg/kg)a 10.38 ± 1.35
Range
Normal range
8–161
65–165
14.47–1181.9 (107–536c)
0.5–5.0
7–63
—
8.33–13.5
—
Data presented as mean ± SD. Data presented as median. Intraquartile range.
a
2 Pediatric Research
b
c
quite variable (range = 14.5–1,182 mU/l), with two cases having TSHdx values above 1,000 mU/l. Most children began treatment within 2 wk of birth (median = 13 d; range = 7–63 d) and their mean ± SD starting dose of L-T4 was 10.38 ± 1.35 µg/ kg, with 71% of dosages above 10 µg/kg/day. Cortical Morphometry Results
Table 3 lists the regions that differentiated CH and TDC groups while Figure 1 provides their CT-difference maps. As per heat bars, regions showing significant thinning in CH (i.e., CHTDC). These results are based on 41 CH cases and 42 TDC since one CH child’s scan was excluded due to excessive motion. Relative to TDC, CH demonstrated significant thinning in four left- and seven right-hemisphere regions. These were located in: (i) frontal and temporal poles and superior frontal gyri bilaterally, (ii) superior parietal gyrus of the left hemisphere, and (iii) middle frontal gyrus and sulcus, inferior temporal gyrus, and precuneus of the right hemisphere. In contrast, CH demonstrated significant thickening relative to TDC in the left central sulcus, supramarginal sulcus, calcarine sulcus, and occipital pole and the right medial orbitofrontal and middle and superior occipital sulci. Most regions showing thinning in CH were gyri, whereas all regions showing thickening were sulci. Effects of Early Disease Indices
Table 4 presents the significant correlations between CT and T4dx while Figure 2 presents brain maps showing significant correlations with T4dx. Figure 3 also shows sample positive and negative correlations for left lateral fissure and right posctentral gyrus respectively. Note that in brain-map figures of correlations, colors are based on heat bars and represent strength and direction of the correlation, not degree of thinning/thickening (as in Figure 1). Since a low T4dx value signifies severe hypothyroidism and a higher value, normalto-near normal TH levels, regions shown in red (positive correlation) indicate the cortex is thinner than normal when hypothyroidism was more severe and regions in blue (negative correlation), a thicker than normal cortex. The following regions were thinner among the CH children with lower T4dx values (positive correlation): superior frontal gyrus bilaterally, cingulate gyrus, lateral fissure posterior segment, and superior temporal gyrus of the left hemisphere, and rectus gyrus and inferior parietal gyrus of the right hemisphere. In contrast, the following regions were thicker when T4dx values were low (negative correlation): left middle frontal sulcus and calcarine sulcus and right postcentral gyrus and inferior temporal sulcus. Table 5 shows opposite effects for TSHdx, which when high signifies severe hypothyroidism. Thus in Figure 4, red indicates regions of increased thickening with more severe hypothyroidism and blue, regions of increased thinning. Figure 5 additionally provides sample correlations for a region showing thickening and one showing thinning. Only Copyright © 2015 International Pediatric Research Foundation, Inc.
Articles
Congenital hypothyroidism and the cortex Table 3. FreeSurfer findings showing regions significantly differentiating CH and control groups in cortical thicknessa Hemisphere
Lobe/region
Structure
Talairach coordinates (x,y,z)
# vertices
Size (mm2)
P value
CH Control Left
Right
Central
Central sulcus
Occipital
Calcarine sulcus
−35, −25, 45
178
57.9
0.0252
−14, −67, 10.5
94
56.3
0.0306
Occipital
Pericalcarine sulcus (occipital pole)
−15, −92, −1
90
75.1
0.0040
Parietal
Inferior parietal (supramarginal sulcus)
−44, −23, 22
138
57.6
0.0264
Frontal
Medial orbitofrontal sulcus
14, 27, −19
110
58.8
0.0334
Occipital
Middle occipital sulcus
28, −83, 7
171
114.3
0.0002
Occipital
Superior occipital sulcus
27, −64, 27
308
144.4
0.0002
CH, congenital hypothyroidism. a Significance is at the P ≤ 0.05 level.
a
c
Central sulcus
Superior occipital sulcus Middle frontal sulcus Rostral middle frontal sulcus
Supramarginal sulcus Middle occipital sulcus
Middle frontal gyrus
Inferior temporal gyrus 5.000
b
Superior parietal gyrus
5.000
1.301
1.301
−1.301
−1.301
−5.000
−5.000
Superior frontal gyrus
d
Superior frontal gyrus
Precuneus
Calcarine sulcus Pericalcarine sulcus
Temporal pole
5.000
5.000
1.301
1.301
−1.301
−1.301
−5.000
−5.000
Figure 1. FreeSurfer (Laboratory for Computational Neuroimaging, Martinos Center for Biomedical Imaging, Charlestown, MA) difference map for cortical thickness. Red tones denote regions where congenital hypothyroidism (CH) participants are thicker relative to typically developing controls (TDC). Blue shades denote regions where CH participants are thinner relative to TDC. Left hemisphere lateral (a) and medial (b) views; right hemisphere lateral (c) and medial (d) views. A P value of ≤0.05 was used for statistical significance. Copyright © 2015 International Pediatric Research Foundation, Inc.
Pediatric Research 3
Articles
Clairman et al.
Table 4. Regions showing significant correlations between T4 at diagnosis and cortical thicknessa Hemisphere
Lobe/region
Talairach coordinates (x,y,z)
Structure
# vertices
Size (mm2)
P value
−15, 19, 52
91
56.6
0.0220
−5, 4, 30
158
56.9
0.0216
T4 positively correlated with CT Left
Right
Frontal
Superior frontal gyrus
Frontal/Parietal
Cingulate gyrus
Occipital/Parietal
Lateral fissure (posterior segment)
−41, −37, 20.5
236
89.3
0.0006
Temporal
Superior temporal gyrus
−51, 5, −17
91
76.5
0.0016
Frontal
Rectus gyrus
7, 51, −16
91
57.5
0.0214
Frontal
Superior frontal gyrus
8, −3, 55
124
65.3
0.0074
Parietal
Inferior parietal gyrus (angular)
41, −60, 43
143
70.5
0.0040
T4 negatively correlated with CT Left Right
Frontal
Middle frontal sulcus
−27, 45, 4
153
89.2
0.0006
Occipital
Calcarine sulcus
−15, −91, 4
73
59.0
0.0158
Parietal
Postcentral gyrus
56.5, −13, 37
211
84.1
0.0008
Temporal
Inferior temporal sulcus
51, −55, 2
102
57.7
0.0210
CH, congenital hypothyroidism; CT, cortical thickness. a Positive correlation indicates lower T4 (more severe CH) at diagnosis leads to thinning of cortical regions listed, negative correlation indicates lower T4 at diagnosis leads to thickening. P value for significance is